public static void Main()
{
string text = File.ReadAllText("../../text.txt").ToLower();
int wordsCount = int.Parse(Console.ReadLine());
string[] wordsOriginal = new string[wordsCount];
string[] wordsLowerCase = new string[wordsCount];
for (int i = 0; i < wordsCount; i++)
{
wordsOriginal[i] = Console.ReadLine();
wordsLowerCase[i] = wordsOriginal[i].ToLower();
}
var occurances = new Dictionary<string, int>(wordsCount);
foreach (string word in wordsLowerCase)
{
occurances[word] = 0;
}
var start = DateTime.Now;
var search = new AhoCorasickSearch();
var matches = search.SearchAll(text, wordsLowerCase);
foreach (var match in matches)
{
string word = match.Match;
occurances[word]++;
}
foreach (var wordOccurances in occurances)
{
Console.WriteLine("{0} -> {1}", wordOccurances.Key, wordOccurances.Value);
}
Console.WriteLine(DateTime.Now - start);
}
public void FindAll_AXBY()
{
var matcher = new AhoCorasickSearch(new List <string>
{
"ax",
"by"
});
matcher.FindAll(" if something then if goto J elsif");
}
private static void searchTest(AhoCorasickSearch search)
{
var res = search.SearchAll("ushersushers");
Console.WriteLine(res);
for (int i = 0; i < res.Length; i++)
{
Console.WriteLine(res[i].ToString());
Console.WriteLine("Match=" + res[i].Match);
}
}
public void FindAll_CommonPostfix_NoWordBreak_NoOverlap()
{
var matcher = new AhoCorasickSearch(new List <string>
{
"if",
"elsif",
}, false);
var actual = matcher.FindAll(" elsif ").ToList();
Assert.AreEqual(1, actual.Count, "Wrong number of matches");
Assert.AreEqual("elsif", actual[0].Value, "Matched wrong keyword matched");
}
static void Main(string[] args)
{
var search = new AhoCorasickSearch();
var keywords = new string[] { "he", "she", "his", "hers" };
search.Build(keywords);
searchTest(search);
searchTest(search);
replaceTest2();
replaceTest3();
Console.Read();
}
public void FindAll_CommonPostfix_NoWordBreak()
{
var matcher = new AhoCorasickSearch(new List <string>
{
"if",
"elsif",
}, false, true);
TextReader rdr = new StringReader("aaa");
var actual = matcher.FindAll(" elsif ").ToList();
Assert.AreEqual(2, actual.Count, "Wrong number of matches");
Assert.AreEqual("elsif", actual[0].Value, "Matched wrong keyword matched (1)");
Assert.AreEqual("if", actual[1].Value, "Matched wrong keyword matched (2)");
}
private static void replaceTest3()
{
var search = new AhoCorasickSearch();
var keywords = new string[] { "一边", "刀锋", "烽火" };
search.Build(keywords);
var text = "一边烽火, 一边烽火天";
var res = search.Replace(text, "-");
Console.WriteLine(res);
text = "一边刀锋很犀利";
res = search.Replace(text, "-");
Console.WriteLine(res);
}
static void Main(string[] args)
{
var search = new AhoCorasickSearch();
var keywords = new string[] { "he", "she", "his", "hers" };
search.Build(keywords);
searchTest(search);
searchTest(search);
replaceTest2();
replaceTest3();
Console.Read();
}
public void FindAll_CommonPostfix()
{
var matcher = new AhoCorasickSearch(new List <string>
{
"if",
"elsif",
});
var actual = matcher.FindAll(" elsif ").ToList();
Assert.AreEqual(1, actual.Count, "Wrong number of matches");
Assert.AreEqual(1, actual[0].Position, "Matched wrong position");
Assert.AreEqual(5, actual[0].Length, "Matched wrong length");
Assert.AreEqual("elsif", actual[0].Value, "Matched wrong keyword");
}
private static void replaceTest3()
{
var search = new AhoCorasickSearch();
var keywords = new string[] { "一边", "刀锋", "烽火" };
search.Build(keywords);
var text = "一边烽火, 一边烽火天";
var res = search.Replace(text, "-");
Console.WriteLine(res);
text = "一边刀锋很犀利";
res = search.Replace(text, "-");
Console.WriteLine(res);
}
public void FindAll_CommonPrefix_NoWordBreakB_NoOverlap()
{
var matcher = new AhoCorasickSearch(new List <string>
{
"if",
"ifa",
"ifb"
}, false);
var actual = matcher.FindAll("xifax").ToList();
Assert.AreEqual(1, actual.Count, "Wrong number of matches");
Assert.AreEqual(1, actual[0].Position, "Matched wrong position");
Assert.AreEqual(3, actual[0].Length, "Matched wrong length");
Assert.AreEqual("ifa", actual[0].Value, "Matched wrong keyword");
}
public FileProcessor(string searchedWord, IDispatcher dispatcher)
{
Dispatcher = dispatcher;
SearchAlgorithm = new AhoCorasickSearch();
Results = new ResultsProvider()
{
SearchedWord = searchedWord,
MatchingFilenames = ImmutableSortedSet <string> .Empty
};
SearchAlgorithm.AddPattern(Encoding.Unicode.GetBytes(searchedWord));
SearchAlgorithm.AddPattern(Encoding.UTF8.GetBytes(searchedWord));
SearchAlgorithm.AddPattern(Encoding.UTF32.GetBytes(searchedWord));
SearchAlgorithm.Freeze();
}
static void Main(string[] args)
{
var search = new AhoCorasickSearch()
{
IsCaseSensitive = true
};
var keywords = new string[] { "he", "She", "hIs", "herS" };
search.Build(keywords);
searchTest(search);
replaceTest(search);
replaceTest2();
replaceTest3();
Console.Read();
}
private static void replaceTest(AhoCorasickSearch search)
{
var text = "ushers";
var res = search.Replace(text, "-");
Console.WriteLine(res);
text = "shersx";
res = search.Replace(text, "-");
Console.WriteLine(res);
text = "her";
res = search.Replace(text, "-");
Console.WriteLine(res);
text = "she";
res = search.Replace(text, "-");
Console.WriteLine(res);
}
private static void replaceTest(AhoCorasickSearch search)
{
var text = "ushers";
var res = search.Replace(text, "-");
Console.WriteLine(res);
text = "shersx";
res = search.Replace(text, "-");
Console.WriteLine(res);
text = "her";
res = search.Replace(text, "-");
Console.WriteLine(res);
text = "she";
res = search.Replace(text, "-");
Console.WriteLine(res);
}
public void FindAll_MultipleMatches()
{
var matcher = new AhoCorasickSearch(new List <string>
{
"if",
"ifa",
"ifb"
});
var actual = matcher.FindAll("Matches 'ifa', 'ifb' and 'if' in a longer text, Keywords: if, ifa, ifb").ToList();
Assert.AreEqual(6, actual.Count, "Wrong number of matches");
Assert.AreEqual("ifa", actual[0].Value, "Matched wrong keyword");
Assert.AreEqual("ifb", actual[1].Value, "Matched wrong keyword");
Assert.AreEqual("if", actual[2].Value, "Matched wrong keyword");
Assert.AreEqual("if", actual[3].Value, "Matched wrong keyword");
Assert.AreEqual("ifa", actual[4].Value, "Matched wrong keyword");
Assert.AreEqual("ifb", actual[5].Value, "Matched wrong keyword");
}
public static void Main()
{
string text = File.ReadAllText("../../text.txt").ToLower();
int wordsCount = int.Parse(Console.ReadLine());
string[] wordsOriginal = new string[wordsCount];
string[] wordsLowerCase = new string[wordsCount];
for (int i = 0; i < wordsCount; i++)
{
wordsOriginal[i] = Console.ReadLine();
wordsLowerCase[i] = wordsOriginal[i].ToLower();
}
var occurances = new Dictionary <string, int>(wordsCount);
foreach (string word in wordsLowerCase)
{
occurances[word] = 0;
}
var start = DateTime.Now;
var search = new AhoCorasickSearch();
var matches = search.SearchAll(text, wordsLowerCase);
foreach (var match in matches)
{
string word = match.Match;
occurances[word]++;
}
foreach (var wordOccurances in occurances)
{
Console.WriteLine("{0} -> {1}", wordOccurances.Key, wordOccurances.Value);
}
Console.WriteLine(DateTime.Now - start);
}
private static void replaceTest2()
{
var search = new AhoCorasickSearch();
var keywords = new string[] { "伟大","特色主义","公园" };
search.Build(keywords);
var text = "从这里建设伟大的特色主义主题公园";
var res = search.Replace(text, "-");
Console.WriteLine(res);
text = "主题公园";
res = search.Replace(text, "-");
Console.WriteLine(res);
text = "伟大的特色主义主题公园";
res = search.Replace(text, "-");
Console.WriteLine(res);
text = "伟大特色主义公园";
res = search.Replace(text, "-");
Console.WriteLine(res);
}
private static void replaceTest2()
{
var search = new AhoCorasickSearch();
var keywords = new string[] { "伟大", "特色主义", "公园" };
search.Build(keywords);
var text = "从这里建设伟大的特色主义主题公园";
var res = search.Replace(text, "-");
Console.WriteLine(res);
text = "主题公园";
res = search.Replace(text, "-");
Console.WriteLine(res);
text = "伟大的特色主义主题公园";
res = search.Replace(text, "-");
Console.WriteLine(res);
text = "伟大特色主义公园";
res = search.Replace(text, "-");
Console.WriteLine(res);
}
/// <summary>
/// Looks for byte patterns that look like
/// </summary>
/// <param name="addrBegin"></param>
/// <param name="addrEnd"></param>
/// <returns></returns>
public IEnumerable<Address> FindPossibleProcedureEntries(Address addrBegin, Address addrEnd)
{
var h = program.Platform.Heuristics;
if (h.ProcedurePrologs == null || h.ProcedurePrologs.Length == 0)
return new Address[0];
byte[] pattern = h.ProcedurePrologs[0].Bytes;
var search = new AhoCorasickSearch<byte>(new[] { pattern }, true, true);
return search.GetMatchPositions(program.Image.Bytes)
.Select(i => program.Image.BaseAddress + i);
}
private static void searchTest(AhoCorasickSearch search)
{
var res = search.SearchAll("ushers");
Console.WriteLine(res);
}
private static void searchTest(AhoCorasickSearch search)
{
var res = search.SearchAll("ushers");
Console.WriteLine(res);
}
private IEnumerable<Address> FindPossibleProcedureEntries(Address addrBegin, Address addrEnd)
{
var pattern = new byte[] { 0x55, 0x8B, 0xEC }; //$TODO: platform-dependent.
var search = new AhoCorasickSearch<byte>(new[] { pattern }, true, true);
return search.GetMatchPositions(prog.Image.Bytes)
.Select(i => prog.Image.BaseAddress + i);
}
private static void AhoCorasickSearch(List <string> list, string txt)
{
AhoCorasickSearch ahoCorasickSearch = new AhoCorasickSearch(list);
}